It's too soon to be able to empirically evaluate any data about the sandbox. Data on data.stackexchange.com is only updated weekly, so many of the riddles posted to the sandbox won't even be in that data yet.
Subjectively, however, it appears to me that the sandbox is having the desired effect and is working well. From my own perspective, it seems that the quality of riddles on the main site has increased significantly.
It seems that most of the riddles posted to the sandbox are making it through to the main site, but I think the sandbox's effect is to discourage those without a well-thought-out riddle from posting at all. If I realize that my riddle is unlikely to pass muster in the sandbox, I won't bother posting it at all.
I thought that after the first couple of days, people would stop checking the sandbox, but there still appears to be lots of activity, and riddles are getting voted up. We will see how things go as we continue with the sandbox, but so far it looks quite promising.
I think the number of people checking the sandbox is irrelevant, if it is having the desired effect. And there are always people who aren't even aware that meta exists, who will only see the riddle on the main site.
From a quick review of the riddle tag, I would say riddle volume has decreased since the sandbox. We were averaging probably at least 5 riddles a day (one recent day had 20), and we've had 5 riddles posted since the sandbox came into effect. But I would argue that the number of decent riddles has remained fairly constant.
There's my thoughts on the matter. I don't know if they help at all.